Population estimates - components of change


The following is an analysis of the components of change for Minnesota counties. The data is from the U.S. Census Bureau’s Annual Population Estimates.

Each year, the United States Census Bureau produces and publishes estimates of the population for the nation, states, metropolitan and micropolitan statistical areas, counties, state/county equivalents, and Puerto Rico. We estimate the resident population for each year since the most recent decennial census by using measures of population change. The resident population includes all people currently residing in the United States.

With each annual release of population estimates, the Population Estimates Program revises and updates the entire time series of estimates from April 1, 2020 to July 1 of the current year, which we refer to as the vintage year. We use the term “vintage” to denote an entire time series created with a consistent population starting point and methodology. The release of a new vintage of estimates supersedes any previous series and incorporates the most up-to-date input data and methodological improvements.

The population estimates are used for federal funding allocations, as controls for major surveys including the Current Population Survey and the American Community Survey (ACS), for community development, to aid business planning, and as denominators for statistical rates, among many other uses. Overall, the estimates time series from 2010 to 2020 was very accurate, even accounting for ten years of population change. The mean absolute percent error (MAPE), which is the average absolute difference between the final total resident population estimates and 2020 Census counts, was only about 2.9 percent across all counties.

We produce estimates using a cohort-component method, which is derived from the demographic balancing equation:

Population base + Births - Deaths + Migration = Population estimate


Vital Statistics

Vital statistics encompass two of the core components of the demographic equation: births and deaths. We receive data on vital records from the National Center for Health Statistics (NCHS) and, until V2024, the FSCPE. NCHS data are derived from birth and death certificates across the United States. Birth data include date of birth, sex of child, residence of mother, and race and Hispanic origin of both mother and father. Death data include residence, age, sex, race, and Hispanic origin of each decedent, and the date each death occurred.

Net Domestic Migration

The third major component of the balancing equation is migration. Migration can be divided into net domestic migration (NDM) within the United States and net international migration (NIM) between the United States and elsewhere. The Population Estimates Program calculates domestic migration using several data sources and methods depending on the age group in question and the level of characteristic detail required.

For state and county total estimates, we calculate county-specific net domestic migration based on four data sources:

  1. Internal Revenue Service (IRS) tax return data for ages 0 to 64;
  2. Medicare enrollment data from Centers for Medicare and Medicaid Services (CMS) for the population aged 65 and older;
  3. Social Security Administration’s (SSA) Numerical Identification File (NUMIDENT) for all ages; and
  4. Change in the GQ population (described in the group quarters section of this document).

To understand how each component is, and has, contributed to population change, we will split up the components in order to fully explore their contribution.


Population

Let’s first explore what the population trends have been over the last 15 years.


Migration

The charts below provide two pieces of information. The dots provide the net number of individuals that in-migrated (above 0) or out-migration (below 0). The second piece of information is are the trend lines that smooth this data to provide the general trajectory of the data using LOESS stats. LOESS in statistics stands for “Locally Estimated Scatterplot Smoothing,” which is a non-parametric statistical method used to fit a smooth curve through a scatterplot of data points by performing local regressions at different points along the data, giving more weight to points closer to the point being estimated; essentially, it helps visualize the relationship between two variables without assuming a specific functional form between them. 

The trends show a couple of themes;

  1. Minnesota on a whole is struggling with migration trends. It does well internationally with consistent in-migration between 10,000 and 15,000 individuals. However, this growth is entirely wiped out by the lost in net domestic migration. This is especially true since 2015.
  2. Hennepin and Ramsey have struggled, and are continuing to struggle, with domestic migration. The pandemic has not helped it.
  3. The western suburbs along with central I94 corridor have been pretty consistently experiencing a modest net in-migration and have been relatively stable.
  4. Greater MN has experienced a significant turn around in terms of domestic migration. The 2000s to 2010s saw them experiencing a worsening net out-migration but has since turned around.


The chart below shows that if it wasn’t for international migration, Minnesota would be a net exporter of people. Domestic migration is consistently a net out-migration while international migration is between 10,000 and 20,000 net in-migration.



The chart below shows that trends are changing.

Domestic migration

Rural areas of Minnesota were experiencing a growing net out-migration from 2000 to 2010. However, that began to change and by 2017 rural areas were experiencing a net in-migration. The opposite is true for entirely urban areas. From 2000 to 2010, domestic migration was improving to the point where experiencing a net in-migration by 2010. However, that trend reversed and they are now trending down to the point where they are experiencing a net out-migration.

International migration

International migration plays a minor role in rural areas of Minnesota, but is significant in our urban areas, and is consistenyl a net in-migration for these county groups.



Entirely rural The recreational version of this county group has done significantly better in terms of net in-migration.

Town/rural mix Similary to the entirely rural folks, the town/rural mix counties that are recreational have done significantly better in migration.

Urban/town/rural mix REcretaional is doing better.



Planning regions in Greater Minnesota follow the patterns we have seen for rural counties - worsening domestic migration through the 2000s, with slow improvement starting around 2012.



Most EDRs are following the similar pattern except for the following;

  • EDR 1 Northwest: it’s been pretty rough
  • EDR 8 Southwest: it’s been pretty rough



These maps are really interesting. They show a couple of things;

International International migration makes up a pretty small percentage of a county’s total population, but the maps show that more and more counties have a positive net-migration of international individuals.

Domestic This shows that rural counties have improved significantly in net domestic migration. There are more rural counties experiencing a net domestic in-migration of individuals by the 2021-2023 timeframe, and counties that are still experiencing a net out-migration are less severe. Lastly, Hennepin and Ramsey are net out-migration counties now.

Net migration Same as domestic. Very similar. One nuance is that some of the more severe DOMESTIC out-migration counties have less severe NET total migration due to positive international migration.




Natural Change

This has tended to play a more prominent role in population change in Minnesota. Let’s take a look at the numbers and trends.

Minnesota is facing an uphill battle when it comes to natural change. Deaths are increasing while births are decreasing steadily. In the early 2000s, there were typically between 30,000 and 40,000 more births than deaths. Now, Minnesota is only experiencing 14,000 more births than deaths. A decline of -53.3%.



Our most rural counties have experienced more deaths than births over the last couple of decades, and it’s going to get worse. From the early 2000s, there were around 250 more deaths than births in these counties. This decade, it has increase to over 500 more deaths than births.

Town/rural mix counties typically had a bit more births than deaths, but that has completely changed so far this decade and we are now experiencing more deaths than births.

Urban/town/rural mix counties experienced significantly more births than deaths in the arly 2000s, but that has completely disappeared and they are now about even.

The entirely urban counties have experienced significantly more births than deaths. But that has begin to lessen. In the early 2000s, these counties were experiencing about 25,000 more births than deaths but has now shrunk to 15,000.



This is super interesting. So even though recreational counties typically do well in migration, they don’t do as well in natural change. For entirely rural counties, and town/rural mix and urban/town/rural mix counties, there are significantly more deaths than births in these counties and it’s getting worse. For the non-recreational counties, they are a bit more even.



Northwest, Southwest, Northeast

Both Northwest and Southwest have just turned the corner and after decades of experiencing more births than deaths, they are now having more deaths than births. Northeast, on the other hand, turned that corner around 2010.

Central, Seven County Metro, Southeast

These three planning regions continue to experience more births than deaths, but that is beginning to decline as births steadily decrease and deaths steadily increase.



Northwest

EDR 1,2,4,5 have identical trends. They were experiencing more births than deaths until relatively recently when it flipped. These regions are all now experiencing slightly more deaths than births.

Northeast

Same as planning region.

Central

EDR 6E and EDR 7E are all now experiencing more deaths than births. EDR 7W is experiencing significantly more births than deaths.

Seven County Mpls-St. Paul

Same as planning region.

Southwest

EDR 6W has been having nearly identical births and deaths for 20 years but has now entered into an era of more deaths than births.

EDR 8 and 9 did have significantly more births but now is about even.

Southeast

Same as planning region.



The following maps provide a pretty clear picture on the trends in the components of natural change. The percent of annual births has been declining over time (becoming less blue), while deaths as a percent of population are increasing (becoming more red). The resulting natural change has been worsening due to these trends (becoming more orange).



What has largest impact so far?

The next question is what is having the greatest impact on population change currently? This can be hard because not only can population go up or down (negative or positive) but so can all the components. So if there is net out migration, yet natural change is positive and population is increasing, how do you measure the impact of the net out migration?

My first idea is to just create bar charts that breakdown each component as a percentage of the total change. Each component would be positive, thus deaths and births each count equally as a unit of change.

The chart below clearly shows that births has been and continues to be the largest contributor to total unit change in population. However, it is shrinking. There is significant growth in deaths as a share of total unit population change. Migration just isn’t that big of a piece.



Entirely rural

Births and deaths in our entirely rural counties have had a similar share of total unit change in population until about 2015. Since than, deaths has been increasing it’s share of total unit change.

Interestingly, in the early 2000s, domestic migration was the largest contributor to total unit change in population with nearly 40% of individuals disappearing from those counties. That has improved significantly since that time frame.

Town/rural mix

Births was the largest contribution to total unit change in population but has since given way to deaths as the largest share. Migration makes up a pretty small percentage of the total unit change.

Urban/town/rural mix

Births made up nearly 50% of the entire unit change in population but that has decreased to 41%. Deaths has increased since that time frame. Domestic migration has gone up and down, but is now a positive share.

Entirely urban

Births continue to make up a sizable share of total unit change in population. Deaths are increasing though while domestic migration continues to be burdensome as well.



Entirely rural

The recreational counties have had the most significant share of their total population unit change from deaths and is currently 54% of the total change. however, births and domestic migration are also sizable.

The non-recreational counties have seen a large increase in births and deaths as a share of the total population unit change while domestic migration has been slowly becoming less bad.

Town/rural mix

The recreational counties are experiencing death as a larger share of population unit change4 while births has declined. The domestic migration has increased significantly as well.

For non-recreational counties, births and death make up very similar shares of population change, while international migration has increased and the net out-domestic-migration has improved as well.

Urban/town/rural mix

In recreational counties, deaths has stated about the same while births has declined as a share and domestic migration has increased.

In non-recreational counties, births continues to be a major driver in population unit change with death increasing. Domestic migration has improved since 2010 as well.



Northwest, Central, Northeast

These planning regions are beginning to experience more deaths than births as a share of population unit change while domestic migration has improved significantly.

Southwest

Southwest has even births and deaths as a share of population change while international migration has increased and domestic migration has improved.

Southeast, Seven County Metro

Both regions have declining births as a share of population change while domestic migration has not been great. International migration continues to make up a sizable portion, however.



EDR 1, 2, 4, 5

These EDRs have a similar pattern - births are beginning to break even with deaths as a share of population change. Domestic migration has either improved but still net out-migration (EDR 1) or are increasing the share towards positive population change (EDR 2, 4, 5).

EDR 6E, 7w, 7E

EDR 6E and 7E each have even births and deaths as a share of population change. 7E has significantly better migration numbers compared to 6E, although international migration has been very helpful for 6E.

EDR 7W has significantly larger percentage of births. In the early 2000s, domestic migration was a huge driver but not has slowed a lot. Deaths are increasing significantly as a share of population change.

EDR 6W, 8, 9

All three have a declining share of population change going towards births, and increasing deaths as a share of pop change. In addition, domestic migration continues to be a net out, but has improved. International migration contineus to be a larger share as well.



The maps below show a few key themes;

  1. Births are declining as a percentage of total population unit change.
  2. Deaths are increasing as a percentage of total population unit change.
  3. International migration makes up less than 25% of total population change across Minnesota.
  4. Domestic out-migration is becoming less severe across rural Minnesota while also changing to a net in-migration.